There was a modest Twitter war—well, more of a dust-up—last Friday between political journalists Sasha Issenberg and Matthew Bai that nicely illuminated an important debate about such journalism. The dust-up concerned an article that Bai wrote last week encouraging Joe Biden to run for office. According to Bai:
Biden is especially popular in Iowa, where he first campaigned for president in 1988, and where he retains unusually strong ties. (The Clintons, you may recall, have never met with great affection there.) I remember being struck, in 2008, by the regularity with which Iowa Democrats told me that Biden was their second choice and would have been first if they thought he could actually win.
Jonathan Bernstein pointed out that, in fact, Hillary Clinton pulled 29 percent of the caucus vote in Iowa in 2008; Joe Biden got less than one percent. Bill Clinton won the state twice in general elections and didn't compete in the 1992 caucus. Meanwhile, Issenberg linked to this poll showing that only four percent of caucusers named Biden as their second choice.
So where is Bai getting this idea that Biden is hugely popular in Iowa while the Clintons aren't? By talking to them. As Bai tweeted to Issenberg, "You should visit Iowa sometime."
Bai is really not making a good argument for visiting Iowa here. His interviews seem to have led him to believe things that are inconsistent with empirical reality.
It's a mistake to think that we can understand everything we need to know about the Iowa caucus through opinion polls and previous election results.
Yet Iowa is exactly the kind of place we need good political reporters to visit, especially this early in the presidential cycle. Party insiders right now are in the midst of making a key decision about whom to nominate next year. It's a murky process that often occurs below the radar, and an incisive reporter can track some of the discussions and transactions that make a nomination happen. They know whose endorsements and donations matter and whose just follow other people's. And the caucus itself is an odd and poorly understood process that doesn't lend itself very well to polling. Reporters who understand just who caucuses, how they are influenced and influence others, and what kind of outcomes to expect can be extremely valuable there.
So there's definitely some value in having good people on the ground in these key early states. But it's just as important for those people to not be led astray by these conversations. A particularly compelling interview subject might convince us that, say, Joe Biden has a real shot at the nomination, when every other empirical indicator we have—money, endorsements, polling, etc.—should tell us that Hillary Clinton has this nomination sewn up.
The real problem with learning about politics by driving around and talking to voters and attendees at rallies is that we're really not getting a representative sample of Iowans or caucus-goers. That's not necessarily bad, but the information we get from that process shouldn't trump the information we get from more statistically sound surveys and election results. We really need both types of information to understand what's going on. Interviews are great at telling us how people make decisions; surveys are great at telling us just how many people support whom.
It's a mistake to think that we can understand everything we need to know about the Iowa caucus through opinion polls and previous election results. There are lots of processes going on that just don't lend themselves well to that kind of analysis. The answers we get from that may be accurate but they'll be incomplete. But it's an even bigger mistake to think we understand public sentiment after talking with an unrepresentative group of Iowans. That's liable to give us wildly inaccurate results.
Seth Masket writes a weekly column on politics for Pacific Standard.